Aliso: A Journal of Systematic and Evolutionary Botany Volume 8 | Issue 4 Article 10 1976 Tribal Interrelationships and Phylogeny of the Asteraceae Sherwin Carlquist Claremont Graduate University; Rancho Santa Ana Botanic Garden Follow this and additional works at: http://scholarship.claremont.edu/aliso Part of the Botany Commons Recommended Citation Carlquist, Sherwin (1976) "Tribal Interrelationships and Phylogeny of the Asteraceae," Aliso: A Journal of Systematic and Evolutionary Botany: Vol. 8: Iss. 4, Article 10. Available at: http://scholarship.claremont.edu/aliso/vol8/iss4/10 ALISO VoL. 8, No. 4, pp. 465--492 SEPTEMBER 30, 1976 TRIBAL INTERRELATIONSHIPS AND PHYLOGENY OF THE ASTERACEAE SHERWIN CARLQUIST Claremont Graduate School and Rancho Santa Ana Botanic Garden Claremont, California 91711 INTRODUCTION Although smaller families of angiosperms may yield to classification prob­ lems readily, the family Asteraceae has posed unusual difficulties for plant taxonomists and phylogenists. It is either the largest family of angiosperms, as Bentham ( 1873) claimed, or, should Orchidaceae prove to be larger, at least the largest family of dicotyledons. Correlated with its large size are its excellent dispersibility, its explosively rapid evolution, its array of numerous instances of parallel evolution within the family, and its capablity for colo­ nizing and adapting to a wide variety of ecological habitats in a weedy fashion ( Burtt, 1961). The consequences of these characteristics are mani­ fold, and the psychological reaction of the plant taxonomist is appropriately laden with blockages. For example, this "most natural" of all families of angiosperms ( Bentham, 1873) continues to be subjected to splitting into two ( e.g., Cichoriaceae) or more families, although those who do this do not all seem to regard the family as polyphyletic. Segregate families are still ranged beside each other in classification systems. The unity of Asteraceae can be cited on numerous bases, and continues to be emphasized, most notably and recently by Fairbrothers et al. ( 1975 ), who have found sesqui­ terpene lactones throughout the family. The parallelisms pandemic in Asteraceae have led taxonomists to seek one or several conservative characters, often without any basis but intuition, and to deduce lines of relationship and specialization accordingly. Parallel­ isms must not be mistaken for polyphyletic origin of the family. There is no convincing evidence that Asteraceae is a polyphyletic group, but there is evidence that particular characters have shown similar shifts two or more times. For example, all-ligulate heads of flowers have evolved in the mutisiad Glossarion ( Carlquist, 1957 c) as well as the tribe Cichorieae. There is no evidence from fossil pollens yet that the family is earlier than Miocene ( Germeraad, Hopping and Muller, 1968; Muller, 1970). Macro­ fossils appearing to be capitula have been claimed from Oligocene ( see Small, 1919, and Cronquist, 1955), but such a fossil as Paleanthus prob­ lematicus from the Cretaceous of New Jersey (Newberry, 1896) is very likely not a composite but a cycadeoid inflorescence. Probably no single character within the family is free from parallel [ 465] 466 ALISO [VoL. 8, No. 4 evolution. For example, styles and stigmas like those of Senecioneae may be found in heleniad Heliantheae ( Carlquist, 1956) as well as in other tribes. Fitchia has stigmatic branches fused nearly to their tips, as do most Mu­ tisieae (Carlquist, 1957c), yet Fitchia is clearly a member of Heliantheae ( Carlquist, 1957b). Raylessness has evolved many times in tribes that are undoubtedly primitively radiate, but that does not indicate affinity among the rayless genera. These examples are obvious ones, but more subtle parallelisms occur and must be taken into account. One should not be surprised that taxonomists have avoided grappling with larger groupings in this worldwide family, only a fraction of whose species are known to the most astute botanist. The more one searches for "reliable" or "conservative" characters whereby to delimit tribes or demonstrate phylesis, the more one finds exceptions. The alternatives available to the taxonomist or phylogenist are virtually all untenable. The most rational option that workers have followed is the construction of natural groupings of genera based on a critically sifted analysis of as many characters as possible ( the "sifting" involves perception of parallelisms) . Cassini's ( 1834) division of Asteraceae into tribes has persisted to the pres­ ent day. Obvious to the student who views Cassini's work is his method of discerning fundamentally distinct genera ( distinct by virtue of a number of characters), then ranging around these tribal "types" genera similar in most or many of these features. This process, not unlike that of numerical taxonomy and computer-formed dendrograms, is an obvious starting point. This method fails where parallelisms are not perceived, where genera are transitional between tribes ( and are perhaps, therefore, "non-missing links"), or where salient aberrant characters have been evolved ( e.g., Aclenocaulon or CoultereUa). With respect to structuring phylogenies within Asteraceae, pitfalls are also difficult to avoid. One must remember that characters, not genera or tribes, are primitive. For example, I have been misquoted to the effect that I believe Mutisieae to be the primitive tribe of Asteraceae. I believe that there are a few genei·a of Mutisieae with a large number of characters primi­ tive for the family, but that does not make those genera primitive-they also have some specialized features. Even if those mutisioid genera have numer­ ous primitive features, calling Mutisieae a primitive hi.be is a misconcep­ tion. Yet in more than one phylogenetic treatment of Asteraceae, characters alleged to be primitive are listed, and the discerning reader of those studies can identify the genus that contains all of these features. Thus, a "primitive genus" has been selected, and characters and their modifications read out accordingly. The circular reasoning in this method is obvious. However, if one attempts alternative methods to designate primitive or specialized features, one cannot cite thoroughly reliable criteria. Related families could furnish suggestive criteria-if we were certain of which families were re­ lated to Asteraceae and which characters in those families are primitive. Consensus and tradition are insufficient grounds for acceptance of either taxonomic treatments or phylogenetic sequences. However, lines of evidence such as anatomy, palynology, and cytology are now sufficiently developed so that a new summary can be made. I am presenting my thoughts on the SEPTEMBER 1976] CARLQUIST: PHYLOGENY OF AsTERACEAE 467 family because the past two decades have seen considerable work published on Asteraceae, because I have an interest as a result of the cumulative experience of my own research in the family, and because my esteemed colleague, Dr. Robert F. Thorne, has urged me to summarize my thinking. Indeed, he and I are in close agreement with regard to the conclusions below. I did, in fact, propose a sort of phylogenetic summary of the family ( 1961, pp. 135-140). However, that summary seems to have escaped notice because it is embedded in a book on plant anatomy. GROSS MORPHOLOGICAL CHARACTERS The pollen presentation mechanism: styles.-Bentham ( 1873) quite rightly cites style-branch morphology as tribal characteristics, while noting that exceptions or ranges of character expression occur in virtually all of the tribes. The range in style features within the family as a whole can be summarized as follows ( tribes cited as examples do contain exceptions) : ( 1) Style branches long ( Vernonieae, Cichorieae, Eupatorieae), of medium length ( Astereae, Heliantheae, Senecioneae, Inuleae), or short, branches fused nearly to the tip ( Anthemideae, Arctoteae, Calenduleae, Cardueae, Mutisieae). ( 2) Stigmatic hairs covering the entire inner surf ace of the branches ( Cichorieae, Cardueae, Eupatorieae, Heliantheae, Vernonieae) to restriction of stigmatic hairs to marginal bands ( other genera of Heliantheae; other tribes). ( 3) Pollen-collecting hairs on the outer surface of style branches promi­ nent ( e.g., some Vernonieae) to hairs sparser ( e.g., some Cichorieae). ( 4) Pollen-collecting hairs scattered on the outer surface of the style branches ( Vernonieae, Eupatorieae, Cichorieae) to hairs localized at the tips of branches ( Senecioneae, Anthemideae) or in a ring below the forking of the branches ( Cardueae, Arctoteae) or restricted to the deltoid tips of the branches ( Astereae, Calenduleae, Heleian­ theae ). ( 5) Style branches acuminate ( Cichorieae, Vernonieae) to acute (Aster­ eae, some Heliantheae) or clavate ( Eupatorieae, Inuleae ) rounded ( Arctoteae, Cardueae, some Mutisieae) or blunt ( Anthemideae, Senecioneae). The prevalent view is that the vernoniad-type style is primitive ( e.g., Cron­ quist, 1955), in that it shows the least "modification" or localization of functions on particular portions of the style: long acuminate branches bearing collecting hairs scattered on the outer surface above and below the point of bifurcation, with an even coating of stigmatic hairs on the inner smfaces of the branches. I tend to agree with this. However, I wish to stress two points that have escaped mention in literature on Asteraceae: ( 1) There is undoubtedly reversibility in all of the features cited above, and different style types can coexist in